The use of artificial intelligence in the diagnosis of neonatal sepsis
Dž. Gojak, K. Gvožđar, Z. Hećimović, Alisa Smajović, Ervina Bečić, Amar Deumić, L. Spahić Bećirović, Lejla Gurbeta Pokvić, Almir Badnjević
Abstract
Neonatal sepsis is an infection that occurs in children up to 28 days of age. Bearing in mind that these are newborns, the procedure of diagnosis and then treatment is extremely difficult. By using artificial intelligence, diagnostics can be speeded up and facilitated. A database of 1,000 data was used to develop the artificial neural network, of which 200 were healthy and 800 were sick. The obtained training effect was 98.93%, sensitivity was 98.75%, specificity was 95.50% and accuracy was 98.33% which indicates the importance of artificial intelligence application which is reflected in the ability to predict the course of the disease, early diagnosis and early access to antimicrobial therapy.